Grok's New Image and Video Tools and Your Website or App: A Guide for Law Firms in USA

Direct answer: xAI has made Grok's image and video generation available to all users on X, with Hotshot's text-to-video technology integrating directly into the platform. For law firms in the USA, this means AI-generated visuals are about to flood the same channels you use for marketing, client communication, and reputation management — and it raises real questions about content authenticity, advertising compliance, and what belongs on your firm's own website versus a social feed you don't control.

According to SocialBee's coverage of X product updates in August 2026, xAI has opened Grok's image and video generation tools to the general user base rather than keeping them behind a paid tier or waitlist, and Hotshot's text-to-video model is being folded into X's own generation pipeline. That single move changes the default assumption every professional services firm has operated under: that convincing video and photo-realistic imagery required a production budget, a shoot, or at least a skilled editor. Now anyone posting on X can generate a plausible-looking video clip or image in seconds, describe it in a sentence, and publish it without disclosure. For an industry built on credibility, precision, and the weight of evidence, that shift lands differently than it does for, say, a consumer brand chasing engagement. Law firms in the USA are about to see more AI-generated content in their industry, in their competitors' marketing, and potentially in disputes their own clients bring to them. A precise adoption number for law firm marketing teams specifically isn't publicly available yet — this is a platform-level change, not a legal-industry survey — so the honest approach is to reason from the pattern: when a major platform removes the cost and skill barrier to synthetic media, the industries that depend on trust and verifiable facts feel it first and hardest.

What xAI Actually Changed With Grok's Image and Video Tools

The practical change is narrow but consequential. Previously, Grok's more advanced generation capabilities were gated — available to subscribers, or limited in scope. Opening them to all users on X means the friction between "I have an idea for a visual" and "there is a finished image or video posted publicly" has effectively dropped to zero for a huge number of accounts. Layering in Hotshot's text-to-video engine adds a second capability: not just still images, but short video clips generated from a text prompt, native to the platform where a large share of legal industry commentary, client complaints, competitor marketing, and even purported "evidence" gets shared first.

This is not a hypothetical automation trend or a distant AI research announcement — it is a distribution decision by a platform with a large, active user base. That distinction matters for how seriously a law firm should take it. A model existing in a research paper is not the same as a model button sitting inside the app your prospective clients, opposing counsel, and journalists already have open. When generation tools move from "available if you seek them out" to "one tap away in the app you already use," usage patterns change immediately, not gradually.

Why the Timing Matters for Professional Services

Professional services firms — law firms especially — tend to adopt visible AI tools slower than consumer brands, and for good reason: the downside of a mistake is reputational and sometimes disciplinary, not just a bad ad. But slower adoption of the tools themselves doesn't mean slower exposure to their effects. Your firm doesn't need to generate a single AI image to be affected by this update. Competing firms will use it in ads. Clients will bring you AI-generated content as "evidence" of harassment, fraud, or defamation. Reviewers and commentators will use it to create content about your firm, accurate or not. The trend is real regardless of whether your firm opts in.

It also helps to be precise about what has not changed. Grok's tools becoming widely available doesn't make synthetic media detection impossible, doesn't override any existing evidentiary rule, and doesn't change what your firm is permitted to say in an ad. What it changes is volume and ease of access — the number of people who can produce convincing synthetic visuals in the time it takes to type a sentence goes from a small technical minority to essentially anyone with an X account. Volume is what turns a theoretical risk into a practical, day-to-day one. A rare event you can handle case by case; a routine one needs a process.

Why This Matters for Law Firms in the USA Right Now

Law firm marketing in the USA already operates under tighter constraints than most industries — state bar advertising rules, restrictions on claims of specialization, disclosure requirements around client testimonials, and in many states specific rules about what a firm can say in an ad without it being misleading. AI-generated imagery and video intersect with almost every one of those constraints at once.

Consider a firm that wants to use Grok's newly opened tools to produce a quick video for a practice-area landing page or a social post — a dramatized "day in the life of a case" clip, or an illustrative image for a blog post about a settlement type. That content did not exist, was not filmed, and depicts no real client or event. Several state bars have advertising rules requiring that dramatizations be clearly labeled as such, and some require disclaimers when a "client" shown is an actor or a scenario is fictionalized. A tool that makes synthetic video trivially easy to produce also makes it trivially easy to accidentally cross a compliance line, because the barrier used to be effort — now it's judgment.

There's a second, sharper concern specific to litigation practices: AI-generated video and images are starting to show up as purported evidence, in insurance claims, employment disputes, and consumer complaints. A law firm's own website and intake process are often the first place a client uploads a photo or video they believe supports their case. With generation this accessible, the odds that some of what comes through intake is partially or wholly synthetic — intentionally or through confusion about a source — go up. Firms that haven't updated their intake workflows to ask pointed questions about media provenance are exposed to that risk without realizing it.

The Marketing Opportunity Is Real Too

It would be incomplete to treat this only as a risk. Firms that market personal injury, family law, immigration, or consumer protection services to a broad public audience have historically needed real production budgets to create explainer videos, testimonial-style content, or visual content for landing pages. Tools like Grok's now make short, clearly-labeled illustrative content — an animated explainer of how a settlement timeline works, a visual walkthrough of what to expect at a deposition — cheap to produce. Used transparently and labeled honestly, this closes a production gap that has favored large national firms with in-house creative teams over smaller regional practices.

What Changes in Practice for Your Law Firm's Website and App

The platform-level change at X doesn't automatically change anything on your own domain — but it should prompt three concrete reviews.

First, your content provenance policy. If your firm's website or client portal accepts uploaded images or video (intake forms, case document portals, client communication apps), you need an explicit, written policy on how staff verify what's submitted, and ideally metadata checks built into the workflow rather than left to individual judgment. This is the same category of problem that identity verification systems solve in other industries — the logic behind Biometric Authentication in Mobile Apps: Face ID, Fingerprint, and Beyond is directly relevant here: verifying that content or a submission genuinely originates from who claims to have created it is becoming as important as verifying identity itself, and the two problems increasingly share technical solutions.

Second, your own marketing content pipeline. If your marketing team or outside partner is going to use AI generation tools for any visual content — even something as simple as a stock-style hero image for a practice page — that decision needs a sign-off step, a labeling standard, and a record of what was generated versus what was photographed or filmed. This is a governance problem before it's a creative one, and it belongs on your website's content management process, not left to whoever happens to be posting that week.

Third, your internal review dashboards. Firms handling higher volumes of client-submitted media, or running active ad campaigns across channels, benefit from a single view that shows what's been submitted, what's been reviewed, and what's flagged for verification. The same thinking that goes into making complex operational data legible applies directly — see Dashboard Design Principles: Making Complex Data Easy to Scan for the underlying approach to surfacing exactly this kind of status information without burying reviewers in noise.

None of these changes require rebuilding your website from scratch. They require deliberate additions to how your site and client-facing app handle media — upload validation, review queues, labeling conventions, and clear internal ownership of the decision to use AI-generated visuals publicly.

Is AI-Generated Video and Imagery a Real Risk for Law Firm Marketing?

Yes, and the risk profile has two distinct layers worth separating.

The first layer is compliance risk on content your firm controls. If your firm publishes an AI-generated video or image without appropriate labeling and it falls under a state bar's dramatization or testimonial disclosure rules, that's a straightforward, fixable problem — update your review process, add the disclaimer, train whoever approves marketing content. This is manageable with process, not technology.

The second layer is harder: content your firm does not control but that affects your firm. A negative review paired with a fabricated "video testimonial." A competitor's ad using a dramatized scenario that implies outcomes your firm can't match without saying so directly. A former client posting AI-generated content that misrepresents interactions with your team. This layer requires monitoring, not just internal policy, and it's worth having someone — internally or through a partner — watching for AI-generated content that references your firm by name, the same way you'd already monitor for fake reviews or impersonation.

Where Website Architecture Comes In

A firm's own website is, in a strange way, the most defensible asset in this environment. Content you host, control, and can verify the provenance of is inherently more trustworthy than content floating on a social platform where synthetic media is now one tap away to create. This is a strong argument for investing more, not less, in your owned web presence — a properly built site with clear authorship, verifiable case results, and a transparent process for how client stories are represented does more to build trust in 2026 than it did five years ago, precisely because synthetic content elsewhere has made verified, owned content more valuable by comparison. This is squarely a Web Development question: how your site structures trust signals, handles media, and presents verified information is now a differentiator, not a formality.

Building a Practical Response: What to Do About It

Start with an audit, not a purchase. Walk through every place your firm's website or app touches media — intake forms, client portals, marketing pages, blog content, social embeds — and note where AI-generated content could enter, intentionally or not. From there:

1. Write a one-page internal policy covering when AI-generated visuals are acceptable in marketing, who approves them, and how they must be labeled to satisfy your state's advertising rules.
2. Add provenance checks to intake. If clients upload photos or video as part of a case, ask directly about the source and consider basic metadata review before that media becomes part of your file.
3. Build a review dashboard, even a simple one, that gives whoever manages your website visibility into flagged or pending media — borrowing the same clarity principles used in other industries' operational dashboards.
4. Monitor mentions of your firm for AI-generated content you don't control, the same way you'd track a fake review campaign.
5. Invest in your owned site as the trust anchor. As synthetic content spreads elsewhere, a well-built, transparent, verifiably authored website becomes a stronger differentiator, not a nice-to-have.

If your firm is also in the process of evaluating outside technology partners for any of this — website updates, an intake portal, a client-facing app — the vetting discipline matters as much as the trend itself. The kind of due diligence questions worth asking a development partner mirror what Insurance Software Development Company: What to Look For Before You Sign lays out for a similarly compliance-heavy, trust-dependent industry: ask about data handling, ask about who actually builds the work versus who sells it, and ask for specifics rather than general assurances.

A Note on Pace

Resist the urge to treat this as urgent-emergency territory. Nothing about Grok's rollout requires your firm to publish anything by next week. What it does require is that the policy and workflow gaps get closed before a mistake happens rather than after — the cost of a proactive one-page policy is trivial compared to the cost of an advertising complaint or a compromised piece of "evidence" that made it through intake unchecked.

It's also worth assigning clear ownership before you start drafting anything. In most firms, marketing content decisions sit with a managing partner or marketing director, while intake and case-file decisions sit with practice group leads or an intake coordinator. AI-generated content policy touches both, and it tends to fall through the cracks precisely because it doesn't fit neatly under either owner's existing job description. Name one person responsible for the marketing-side policy and one for the intake-side process, even if it's the same person wearing two hats, so the audit described above actually gets finished instead of stalling as everyone assumes someone else is handling it.

What This Work Typically Costs

Most of what's described above — provenance checks on intake, a review dashboard, updated content workflows on your website, and stronger trust signals across your site — falls under standard web development engagement tiers rather than a separate "AI risk" line item.

Tier	Typical scope	Fits this scenario when
Essential — $1,000	Focused fixes: intake form updates, labeling/disclosure additions to existing pages, basic upload validation	You need the provenance and labeling gaps closed on an existing site without a broader rebuild
Growth — $2,000	A review dashboard for flagged media, expanded intake workflow, stronger trust-signal design across key pages	You're actively marketing with visual content and need ongoing internal review tooling
Enterprise — $4,000+	Full client portal or intake system rebuild with provenance verification, multi-office review workflows, and integrated monitoring	Your firm handles high submission volumes across multiple practice areas or locations

These are starting points reflecting typical scope at each tier, not fixed quotes — the right tier depends on how much of your current site and intake process already supports structured review.

Key Takeaways

• xAI opening Grok's image and video tools to all X users, with Hotshot's text-to-video integration, removes the cost and skill barrier that previously limited synthetic media — law firms will encounter it whether or not they use it themselves.
• The two real risks are compliance (unlabeled AI-generated marketing content triggering state bar dramatization or disclosure rules) and evidentiary (synthetic media entering client intake as purported proof).
• Your firm's own website and intake process are the most controllable point of defense — add explicit provenance checks and a one-page internal policy on AI-generated marketing content before you need one.
• A simple internal dashboard for reviewing flagged or pending client-submitted media pays for itself the first time it catches something that shouldn't have gone further unverified.
• Owned, verifiably authored web content becomes more valuable, not less, as synthetic media spreads elsewhere — this is a reason to invest in your site's trust signals, not defer it.
• Vet any development partner on data handling and provenance specifics the same way you'd vet any compliance-adjacent vendor.

If your firm needs help auditing where AI-generated media could enter your website or client intake, or wants a practical review workflow built into your site rather than left to individual judgment, book a meeting with our team.

Frequently Asked Questions

What exactly did xAI change with Grok's image and video tools in August 2026?

xAI opened Grok's image and video generation capabilities to all users on X rather than restricting them to a paid tier or limited group, and is integrating Hotshot's text-to-video technology directly into the platform. This means generating a photo-realistic image or short video clip from a text prompt is now available to essentially anyone using X, according to SocialBee's coverage of the update.

Can AI-generated video count as evidence in a legal case?

Courts are actively grappling with this question, and standards vary by jurisdiction and case type. What's clear is that as generation tools become more accessible, the burden on firms to verify the provenance of client-submitted media before relying on it increases, making intake-stage verification more important than it was even a year ago.

What state bar advertising rules apply to AI-generated marketing content?

Most states require that dramatizations, actors portraying clients, and fictionalized scenarios in legal advertising be clearly disclosed, and several require specific disclaimer language. AI-generated video or imagery used in marketing typically falls under these same dramatization disclosure rules, so firms should treat it identically to a scripted commercial rather than as a lower-scrutiny category.

Is it unethical for a law firm to use AI-generated images in marketing?

Not inherently — the ethical issue is misrepresentation, not the generation method itself. An AI-generated illustrative graphic clearly presented as such is different from AI-generated content designed to look like a real client testimonial or case outcome without disclosure.

How can my firm tell if a client-submitted photo or video is AI-generated?

There's no single reliable test, but basic steps include checking file metadata for inconsistencies, asking direct questions about the media's origin during intake, and treating any media that a client can't clearly explain the source of with added scrutiny before it's relied upon.

Should my firm add a specific question about AI-generated content to intake forms?

Yes, this is a low-cost, high-value addition. A direct question asking whether any submitted photos or videos were edited, generated, or altered by AI tools creates a paper trail and prompts clients to disclose things they might not otherwise think to mention.

What's the difference between Grok's tools and other AI image generators for marketing purposes?

For a law firm's purposes, the specific generator matters less than the fact that generation is now embedded directly into a major social platform your clients and prospects already use. The compliance and authenticity questions are the same regardless of which underlying model produced the content.

Will opposing counsel start using AI-generated content against my clients?

It's a reasonable risk to prepare for rather than assume away. Any firm handling disputes involving photo or video evidence should build provenance verification into its process now rather than reactively after encountering a disputed piece of media in a live case.

How much does it cost to add media provenance checks to a law firm website?

Focused additions like upload validation and updated intake questions typically fall in the Essential tier around $1,000, while a full review dashboard and expanded workflow moves into the Growth tier around $2,000, depending on how much of your current intake system already supports structured checks.

How long does it take to build a review dashboard for flagged client media?

A focused dashboard covering flagged-media review, typically scoped at the Growth tier, usually takes a few weeks from requirements to launch, depending on how it needs to integrate with your existing intake or case management system.

Can this be added to our existing website, or do we need a full rebuild?

Most firms can add provenance checks, labeling standards, and a review dashboard onto an existing site without a full rebuild. A full rebuild is typically only necessary if your current intake system has no structured data capture at all.

What should our internal AI-generated content policy actually say?

At minimum, it should state when AI-generated visuals are permitted in marketing, who signs off before publication, how the content must be labeled to satisfy applicable state bar rules, and where records of what was generated are kept for compliance purposes.

Are there specific risks for personal injury firms versus other practice areas?

Personal injury and consumer protection firms tend to rely more heavily on visual marketing (explainer videos, illustrative scenarios) and on client-submitted photo/video evidence, so both the marketing compliance risk and the intake verification risk are more concentrated for these practice areas than for, say, transactional practices.

Does this affect law firm apps differently than websites?

Client-facing apps that accept media uploads face the same provenance questions as websites, but often with less scrutiny applied because app interactions can feel more "internal" or trusted. The same verification standards should apply regardless of channel.

What is Hotshot, and why does its integration into X matter?

Hotshot is a text-to-video generation technology that xAI is integrating into Grok's capabilities on X. Its relevance to law firms is that it adds native, in-platform video generation to a channel already used heavily for legal marketing, client commentary, and public dispute airing.

Should our firm monitor social media for AI-generated content that mentions us?

Yes, particularly if your firm has an active public profile or handles high-visibility cases. Monitoring for AI-generated impersonation or misleading content that references your firm follows the same logic as monitoring for fake reviews, and catching it early limits reputational damage.

Can AI-generated marketing content help smaller regional firms compete with national ones?

Used transparently and labeled correctly, yes — the production cost gap that historically favored large firms with in-house creative teams narrows when illustrative content becomes cheap to produce. The compliance discipline required to use it correctly matters more than the firm's size.

What happens if our firm accidentally publishes unlabeled AI-generated content?

Depending on your state bar's specific rules, this could range from a required correction to a formal advertising complaint. The practical fix is catching it before publication through a sign-off step, which is far cheaper than remediation after the fact.

Is there a legal requirement to disclose that a website image was AI-generated?

Requirements vary by state and by context — advertising-specific disclosure rules are more established than general website content rules. When in doubt, treating AI-generated imagery the way you'd treat a stock photo or dramatization, with appropriate context, is a reasonable default.

How do we train staff to spot potential AI-generated content in client submissions?

Basic training should cover common inconsistency signs (odd lighting, mismatched shadows, unnatural motion in video, metadata gaps) alongside a simple rule: anything ambiguous gets escalated for a direct conversation with the client about its source rather than being accepted at face value.

Will courts eventually require disclosure of AI-generated evidence automatically?

Some jurisdictions are moving toward specific evidentiary rules around AI-generated or AI-altered media, though standards are still developing nationally. Firms are better positioned by building their own verification habits now than by waiting for uniform court rules to catch up.

What's the single highest-priority action for a law firm right now?

Auditing where media enters your website or app — intake forms, portals, marketing uploads — and adding a basic provenance question or check at each of those points is the highest-leverage, lowest-cost first step.

Does Scult build custom intake portals for law firms?

Yes, through Web Development engagements scoped to a firm's specific intake and case management needs, ranging from focused additions to existing sites up to full custom portal builds.

How does dashboard design relate to reviewing AI-generated content?

A well-designed review dashboard makes it fast for staff to see what media has been submitted, what's flagged for review, and what's been cleared — the same scanability principles that apply to any operational dashboard apply directly to a media review queue.

What's the connection between biometric authentication and content provenance?

Both problems are fundamentally about verifying that something (an identity, a piece of content) genuinely originates from where it claims to. The verification techniques used for identity — layered checks rather than a single signal — translate directly to verifying media provenance.

Should our firm avoid AI-generated content entirely to stay safe?

Avoidance isn't necessary or particularly protective — the risk comes from unlabeled or undisclosed use, not from use itself. A firm with a clear labeling policy can use AI-generated illustrative content safely; a firm without any policy is exposed whether it uses the tools or not.

How do we handle AI-generated content in client testimonials specifically?

Testimonials require particular care because state bar rules on client testimonials are typically stricter than general advertising rules. Any AI-generated element in testimonial content — a dramatized voice, a generated visual — should be clearly disclosed as such, not left ambiguous.

What's a reasonable timeline for implementing these changes across our firm?

A basic policy and intake question addition can be in place within a week or two. A review dashboard and expanded workflow typically takes a few weeks. Treat it as a near-term priority rather than an emergency, but don't let it sit indefinitely.

Does this apply to solo practitioners and small firms, or only large firms?

It applies across firm sizes. Solo practitioners and small firms often have fewer internal checks in place, which can make a simple provenance policy and intake question even higher-value relative to their size.

How do we know if our current website already handles this adequately?

Walk through every point media enters your site or app and ask whether there's a documented check at that point. If the answer is "we'd just notice if something looked off," that's a gap worth closing with a structured process instead.

What role does content labeling play in avoiding compliance issues?

Clear labeling is often the single factor that separates compliant use of AI-generated marketing content from a bar complaint. A labeled dramatization is standard practice; an unlabeled one presented as real is the actual risk.

Can AI-generated video be used for law firm explainer content on YouTube or social media?

Yes, provided it's clearly presented as illustrative or animated content rather than depicting real events or real clients without disclosure. This is one of the more straightforward, lower-risk uses of the newly accessible tools.

What should we ask a development partner about handling sensitive client media?

Ask specifically how uploaded media is stored, who has access, whether metadata is preserved for verification purposes, and how the vetting process works — the same category of due-diligence questions outlined for evaluating a partner in Insurance Software Development Company: What to Look For Before You Sign applies directly here.

Is there a difference between AI-generated content risk for B2B legal services versus consumer-facing firms?

Consumer-facing firms (personal injury, family law, immigration) face more exposure because of higher marketing visibility and higher intake volume of client-submitted media. B2B-focused firms face more exposure on the marketing compliance side than on intake risk.

How often should our AI content policy be reviewed and updated?

Given how quickly platform capabilities are changing, a light review every two to three months is reasonable — not a full rewrite, but a check that the policy still reflects what tools are actually accessible and in use.

What's the cost difference between a basic labeling fix and a full provenance system?

A basic labeling and disclosure fix on existing marketing pages typically falls in the Essential tier around $1,000. A full provenance verification system integrated into intake, at the Enterprise tier, starts around $4,000 depending on scope and integration complexity.

Does our firm need a lawyer to review our AI content policy, or is this a marketing/IT decision?

Both. The policy's substance (what disclosure is required, what's compliant under your state's rules) needs legal review; the implementation (how it's enforced on your website and in your workflows) is a technical and marketing execution question.

What happens to firms that ignore this trend entirely?

Nothing forces immediate action, but the risk compounds quietly — an unlabeled AI-generated ad, an unverified piece of "evidence" accepted at intake, or a fabricated online complaint about the firm are all more likely the longer no policy or check exists.

Are there tools that automatically detect AI-generated images and video?

Detection tools exist and are improving, but none are fully reliable, and detection accuracy varies significantly by content type and generation method. Treat automated detection as one input alongside direct client conversation and metadata review, not a standalone solution.

How does this trend affect law firm SEO and content marketing strategy?

As synthetic content spreads on social platforms, owned website content with clear authorship and verifiable information becomes a stronger trust and ranking signal by comparison — reinforcing the value of investing in your own site's content quality rather than relying solely on social distribution.

Should we update our client engagement letters to address AI-generated evidence?

It's worth considering language that asks clients to disclose whether submitted media was AI-generated or altered, giving your firm a documented basis for later verification questions if the media's authenticity becomes relevant.

What's the risk of a competitor firm using misleading AI-generated ads?

Beyond the ethical question for that firm, it can create pressure to match perceived production quality. Resist matching questionable practices — a well-labeled, transparent alternative is a differentiator, not a disadvantage, especially as awareness of synthetic marketing content grows.

Can AI-generated content help with multilingual marketing for immigration or consumer firms?

Potentially, for illustrative graphics or explainer visuals adapted across languages, provided the same labeling and disclosure standards apply regardless of the language the content is published in.

How should our firm respond if a client uploads what turns out to be AI-generated "evidence"?

Address it directly and promptly with the client rather than treating it as a minor technicality — depending on the case, it can affect strategy, credibility, and disclosure obligations. Having already asked about media origin at intake makes this conversation far less awkward when it happens.

What's the realistic timeline before AI-generated video becomes indistinguishable from real footage?

Detection is already difficult in many cases and continuing to get harder as generation quality improves. This is precisely why provenance verification processes matter more than after-the-fact detection — asking the right questions upfront is more reliable than analyzing content after it's already in your file.

Does our firm's cyber insurance or malpractice coverage address AI-generated content risk?

Coverage specifics vary significantly by policy and carrier, and this is a question worth raising directly with your insurance broker or carrier rather than assuming existing coverage addresses it — this is an emerging risk category many policies weren't written with in mind.

What's a realistic first project scope for a firm just starting on this?

A focused Essential-tier engagement — updated intake questions, basic upload validation, and a labeling policy for existing marketing pages — is a reasonable, low-risk starting point before considering a larger dashboard or portal investment.

How do we measure whether these changes are actually working?

Track how many flagged media submissions your review process catches over time, whether any marketing content has needed correction for disclosure issues, and whether staff can consistently answer "how do we know this is real" for content on your own channels.

Will AI-generated content regulations tighten for legal marketing specifically?

It's a reasonable expectation given the pace of platform changes, though specific legal-industry rules haven't caught up uniformly across states as of mid-2026. Firms that build sound internal practices now will adjust more easily than firms waiting for regulation to force the issue.

Where should a law firm start if this is the first time addressing AI-generated content risk?

Start with the audit described in this guide — walk your website and app for every point media enters, write the one-page policy, and add a single provenance question to intake. From there, book a meeting if you want help scoping the technical pieces.