NEWThe Value Intelligence Buyer’s Guide 2026.Read the guide →

Build Your Own Business Case AI? Build the Workflow, Not the Value Model

You can build business cases with ChatGPT, Claude or Copilot, and many teams do. The assistant drafts fast, but it doesn't supply the value model behind the numbers. For most teams the answer is both: build the workflow in the assistant you already use, and connect it to a value model valueIQ generates, through its MCP server.

Published by valueIQ, one of the options compared, and judged by the same rule as every other. Every fact about building it yourself and about valueIQ comes from public pages, read 23 to 29 October 2026 for the Value Intelligence Buyer's Guide, where "built yourself" is one of the eight options. Corrections: research@valueiq.ai.

Facts checked 23 to · Last reviewed

Can you build business cases with ChatGPT or Claude?

Yes. A general-purpose AI assistant can research a customer, structure a business case and format it for an executive, and the model vendors now publish business case tooling themselves. Anthropic's open-source sales plugin includes a skill that sets price against quantified value on each deal (buyer's guide), and one Anthropic customer story describes an ROI calculator built in 60 minutes (OpusClip).

Teams that build it themselves usually do it in one of four ways, and many run more than one at once:

Can you build business cases with ChatGPT or Claude?
How teams build itWhat it isWho keeps it current
Ad-hoc promptingA seller pastes discovery notes into an assistant and asks for an ROI estimate or a business caseEach seller, deal by deal
A shared assistant with a templateA Claude Project, a ChatGPT workspace, a Gemini Gem or a Copilot agent loaded with the team's template, case studies and pricingWhoever writes the instructions and the knowledge files
A skill plus a spreadsheetA packaged method, such as a Claude skill, that drafts the case, with the arithmetic in an Excel or Sheets modelWhoever owns the method and the spreadsheet
An internal applicationAn app built on the model APIs, with retrieval over case studies and benchmarks and a CRM connectionRevOps or engineering, with evaluation and security review

What does building it yourself do well?

Building it yourself is quick to start, uses tools the team already has, and keeps the method in the team's hands. On the public record:

  • No new vendor. Claude and ChatGPT have free plans, and Claude Team and ChatGPT Business cost $20 a seat a month, billed annually. Team plans are self-serve, with no services needed (Claude pricing).
  • Drafts in your own templates. Anthropic's published sales skills produce decks, one-pagers, documents, spreadsheets and web pages in the team's own templates (buyer's guide).
  • Your stack, connected. Claude connects to Salesforce (in beta) and HubSpot, Gong and Fireflies feed call transcripts into the proposal, and an admin can enable a Gainsight connector (Claude for sales; Gainsight connector).
  • Governance you control. Owners approve skills and plugins before they are published to the organization (buyer's guide).
  • Your data terms. Business AI plans, such as Microsoft 365 Copilot, state that prompts and company data aren't used to train models (buyer's guide).

Best for: a first experiment, a small team, or a seller who needs a draft today, with someone on the team to write and maintain the prompts and templates.

Not the right fit for: teams that need every seller to use the same method and benchmarks. The team writes and maintains the prompts, and benchmarks come from web search at request time, so two sellers can get two different cases.

Where does a homegrown business case need a value model behind it?

A homegrown business case needs a value model the moment finance asks where the numbers came from. The assistant works from the context it is given: it supplies the method, the benchmarks and the assumptions itself, so two sellers get two different cases and finance can't trace the numbers (buyer's guide FAQ).

Four questions decide whether a case holds up, whoever builds it:

  1. Where does each number come from? A figure drawn from the prompt and the model's general knowledge can't be traced to a source.
  2. Would two sellers get the same answer? Each person pastes different context, so the same deal can produce two cases.
  3. Has the customer confirmed the inputs? An assumption the buyer never confirms is still a guess.
  4. What is the next-best alternative worth? Economic value is measured against what the customer would otherwise do, including the competitor in the deal, not as a return on a list of features.

The answer is a value model: the value drivers, the equation behind each one, the variables that feed it, and market benchmarks with their sources. Writing and maintaining one by hand, for every product, segment and competitor, is the work a trained value engineer spends days, often weeks, on. That is the part worth not building yourself.

How does building it yourself compare with valueIQ?

The table compares building it yourself with valueIQ on the buyer's guide rows that decide whether a case holds up. Each cell links to its source; where the source is help documentation, the link goes to the guide's feature table, which links the original.

valueIQ’s column is shaded so you can find it, not ticked.
What the buyer's guide checkedvalueIQBuilt yourself
What has to exist before the first caseCompany website and product name; for a deal, the prospect's website (source)An AI subscription, plus the team's own prompts, templates and source files (source)
Who builds the model, and who builds each casevalueIQ generates the model, your team signs it off; AEs build each case alone (source)The buyer's own team writes and maintains prompts; any seller drafts a case alone (source)
Where benchmarks come fromMarket benchmarks researched per model, each listing its sources with links (source)Web search at request time; no benchmark library published (source)
Every number shows where it came from Yes. Each estimate lists sources with links; provenance state per variable (source) Partial. Published skill cites the source of each value; no record of who set it (source)
The buyer's own numbers replace estimates Yes. Invited buyer edits inputs and confirms them; confirmations kept with history (source) Partial. Buyer numbers enter from transcripts, email and CRM; unevidenced claims flagged (source)
Value against a named competitor on this deal Yes. Competitor model attached to the deal, valued on the same customer variables (source) Not found. Competitor research and battlecards on request, no competitor valuation (source)
Price as a share of value, across all deals Yes. Value Capture on every deal and across the pipeline, with trend and segment (source) Partial. Business case sets price against value per deal; no cross-deal ratio view (source)
Works with AI assistants and agents (MCP) Yes. Remote MCP server live, included on every plan including Free (source) Not found. Assistants connect to MCP servers but publish none for cases (source)
Try it without a sales call Yes. Free plan, 500 credits a month, public-data models, no integrations (source) Yes. Free plans on Claude and ChatGPT; trials on Copilot, Workspace, Gemini Enterprise (source)
Prices published Yes. Free; Professional $120 a month; Teams $960 a month; Enterprise custom (source) Yes. Claude Team and ChatGPT Business $20 a seat a month, billed annually (source)

"Not found" means we couldn't find it on the public record after checking twice, not that the capability doesn't exist. All 41 rows, across all eight options, are in the buyer's guide feature table.

Build, buy or both?

Most teams don't have to choose. Build the workflow in your assistant, and let it call a value model it didn't have to invent.

Build, buy or both?
Build it yourselfvalueIQ on its ownBoth: your assistant plus valueIQ
The methodYour team writes it into prompts, templates and skillsGenerated by specialized agents trained in economic value estimation; your team signs off the modelYour skill supplies your steps and review rules; valueIQ supplies the value model
The benchmarksWeb search at request timeMarket benchmarks researched for each model, each with its sourcesvalueIQ's, read through MCP
Every seller, every dealDepends on each seller's prompt and contextOne model behind every caseOne model behind every workflow and every assistant
The competitorResearch and battle cards on requestCompetitor valued on the same customer's numbers, with its pricing page analysedYour agent pulls the competitor's pricing analysis and compares the two models driver by driver
Where the work happensYour assistantvalueIQ's appYour assistant, through the MCP server
What it costsThe AI seats you already pay for, plus your team's time to build and maintain itFree plan; Professional $120 a month; Teams $960 a month; Enterprise customYour AI seats plus a valueIQ plan; reading deals, coaching and drafting through MCP are free, and creating uses credits

Sources: buyer's guide; Value Intelligence for AI; valueIQ pricing.

How does valueIQ work inside the assistant you already use?

valueIQ works inside your assistant through its MCP server. Claude, ChatGPT, Cursor, VS Code Copilot and Windsurf connect on supported plans, any MCP-compatible client can connect, and the MCP server is included on every valueIQ plan, including Free (Value Intelligence for AI; MCP).

Four layers do four jobs. Generative AI writes and reasons. A skill supplies the method: the steps, the questions to ask and the review rules. MCP supplies the connection. valueIQ supplies the intelligence: value models, business cases, competitor value models and pricing analysis.

What your assistant can do with valueIQ connected (as of October 2026):

  • Read your deals and business cases
  • Generate a business case for a deal
  • Run deal coaching grounded in the deal's business case
  • Pull competitor pricing analysis
  • Create and update deals
  • Add a competitor value model to a deal
  • Browse workspaces, products and market benchmarks
  • Save and resume coaching across turns

Control stays with you. The assistant connects with OAuth 2.1 and PKCE, every call runs under the signed-in user's own identity and permissions, valueIQ does not use deal data passed through MCP for model training, and nothing sends, presents or publishes on its own (Value Intelligence for AI).

You don't have to stop building. Build the workflow and call the engine. If your team already has a value model or framework of its own, you can put it in and the platform runs on yours (Value Intelligence for AI). Steven Forth sets out the approach in Build Value Intelligence Into the AI Your Team Already Uses.

What valueIQ brings that a prompt can't

valueIQ's value intelligence software generates a value model for any product, at scale. Its specialized agents are trained in economic value estimation, the method pricing professionals use to put a number on what an offer is worth against the customer's alternative. From the product's URL, they build the model in about 25 minutes: the value drivers, the equation behind each one, the variables that feed it, and market benchmarks with their sources. Add your own product material, such as case studies or customer results, and the model gets sharper still. Each business case is then built on it in about five minutes; add what you already know about the deal, such as an RFP or discovery notes, and valueIQ fits the case to that buyer (valueIQ). On every deal, vIQ, valueIQ's copilot, works like a value engineer at the seller's side, grounded in that deal's numbers (vIQ). valueIQ has generated 500+ value models and analysed 2,000+ pricing pages.

  • A value model for every product, from its URL. Every product, segment and competitor gets a value model, kept current, including the ones a team has already modelled.
  • Every number shows where it came from. Each number is marked as a market default, a figure the seller set or a number the customer confirmed, and each estimate lists its sources with links (Evidence).
  • The competitor valued on the same deal. In the buyer's guide, valueIQ is the only option whose public pages show a named competitor valued on the same customer's numbers, with its pricing page analysed beside the case (competitor models).
  • Value connected to price across deals. Value Capture on every deal and across the pipeline, discounting measured against expected value, and pricing approaches tested against value (pricing teams).
  • Open models. valueIQ's own value and pricing models are published as open Value Project files that any AI can read (value model).

Best for: teams that need value models for many products, segments or deals, at a speed people alone can't reach. That means B2B software and AI companies selling on value; value engineering teams that want a model for every product, kept current, and a business case on 100% of the pipeline; and consulting and pricing practices scaling their value research across clients.

Not the right fit for: sales motions that never need a value argument: low-price, self-serve or transactional deals.

Price: Free plan; Professional $120 a month; Teams $960 a month; Enterprise custom (pricing).

How should you decide?

Decide by how often finance checks the case, and how many products, segments and competitors the team sells against.

  • Your team builds agents or skills in Claude or ChatGPT already:build the workflow there, and connect valueIQ through MCP for the value model.
  • Every deal that matters needs a case finance can check, across many products or competitors:valueIQ, in its own app or inside your assistant.
  • One seller, an occasional deal, and cases that rarely face a finance review:an assistant on its own may be all you need.

Start with one workflow and one real deal. Connect the MCP server on the Free plan, test it, and measure whether the case gets faster or stronger.

Frequently asked questions

Can I build a business case with ChatGPT or Claude?

You can draft a business case quickly with a general-purpose assistant, and many teams do. The gap is the model behind it: the assistant supplies the method, the benchmarks and the assumptions itself, so two sellers get two different cases and finance can't trace the numbers. Connecting the assistant to a value model closes that gap.

Is building it yourself cheaper than buying value intelligence software?

The seats are cheap: Claude Team and ChatGPT Business cost $20 a seat a month, and valueIQ starts on a Free plan. The larger cost of building is people: someone has to write the method, find and update the benchmarks, maintain the prompts and integrations, and check the output. That work repeats for every product and segment.

Can valueIQ work inside Claude or ChatGPT?

Yes. Claude, ChatGPT, Cursor, VS Code Copilot and Windsurf connect to valueIQ through its MCP server on supported plans, and the server is included on every valueIQ plan, including Free. Your assistant can read deals, generate business cases, run deal coaching and pull competitor pricing analysis, and nothing is sent without your review.

Do we have to stop building our own AI tools?

No. Build the workflow and call the engine: keep your own prompts, skills and agents, and let them call valueIQ for the value model, the business case and the competitor analysis. If your team already has a value model or framework of its own, you can put it in, and the platform runs on yours.

Is a homegrown AI business case good enough for value selling?

For value selling, a business case has to survive the economic buyer's questions: where each number came from, what the alternative is worth, and whether the customer agreed the inputs. A homegrown case can pass if someone supplies a real method and sourced benchmarks. Read more about value selling.

About this page

valueIQ publishes this page and is one of the options it compares. Facts about building it yourself and about valueIQ come from the Value Intelligence Buyer's Guide (November 2026 edition), which read the public record between 23 and 29 October 2026, and from valueIQ's Value Intelligence for AI page. If a fact is wrong, send the public source to research@valueiq.ai and we'll update the page.

valueIQ is value intelligence software that generates value models for any product, at scale, so B2B companies can define, quantify, price, communicate and prove the value they bring to their customers. From a product's URL, its agents, trained in economic value estimation, build a value model in about 25 minutes, ready for executive-ready business cases, competitor value models and pricing analysis on every deal. Revenue teams, value engineering teams and consulting and pricing practices use it for value selling, value management and pricing. valueIQ is the creator of the valueIQ Value Intelligence Maturity Model™. The Value Project (thevalueproject.org), an open standard for machine-readable value and pricing models, was initiated by valueIQ.