# Titleman — AI engine that turns a commercial real estate brief into finished underwriting, decks and diligence.

> Source: AgentsIndex — https://agentsindex.ai/titleman (structured, researched, re-verified)
> Facts last verified: 2026-10-20

Titleman is an AI engine for commercial real estate analysis that reads a deal's own documents, enriches them with county, tax and listing records, and returns underwriting models, value opinions, comps, decks and diligence findings. Every figure carries its source and date.

| Fact | Value |
| --- | --- |
| Website | https://titleman.ai/ |
| Price | Price on request |
| Free tier | No |

The vendor states 'No trial. A pilot on agreed numbers.' Access opens only once pilot terms are agreed in writing and the pilot fee is settled; the pilot runs 14 days at most and its fee counts toward the annual agreement. The MCP connector's read-only tools cost no credits, but they require an existing account.

## What it does on its own

**Acts on its own** · Partly — It plans and runs a multi-step chain from one brief without step-by-step approval, but the vendor states the output is a draft that a named reviewer on the client's team must check and sign, and a run can stop to ask a person a question.

> Collaborative, not autonomous. Every output carries its sources and a confidence clamped to the weakest one, then goes to a named reviewer on your team.
> — https://titleman.ai/platform · Product page, 'Collaborative, not autonomous' block at the foot of the page

**Several agents** · Yes — One brief is broken into planned steps that run as parallel tracks across chained modules, and the enterprise page talks about where the agents run, so several workers split a single job.

> Titleman works tracks run in parallel ... Brief in plain words Steps planned Modules chained One reconciled output
> — https://titleman.ai/ · Homepage, 'Every section of Titleman' panel, step 2 'Titleman works'

**Limits on what it may do** · Yes — Over the MCP connector read tools are annotated read-only, paid runs return an estimate and start only on explicit confirmation, and a workflow run is capped by a caller-set dollar budget of at most $50.

> it starts with confirm: true and an estimate within your max_usd (at most $50). A step whose estimate would pass the budget is not started
> — https://titleman.ai/docs/claude-connector · Connector documentation, Tools > Write table, run_workflow row

**No code needed** · Yes — The job is briefed in plain words and saved chains are created by saying them once, with the vendor explicitly ruling out a canvas or automation project, so a non-technical analyst can run it.

> No canvas, no automation project — one message, saved as a workflow.
> — https://titleman.ai/platform · Product page, Workflows section subheading

## Which models it runs on

**You pick the model** · Partly — No model picker is documented for ordinary users, but the vendor says which models the agents use is settled with the firm during an enterprise deployment, so the choice exists only inside a negotiated engagement.

> Where the agents run, whose models they use and who answers for them — agreed with your firm.
> — https://titleman.ai/demo · Demo page, 'From one deal to your whole desk', step 02 deployment plan

## How you reach it

**In the browser** · Yes — Titleman is sold as a hosted platform with a sign-in and a web app, and the Platform Agreement defines the Service as software-as-a-service delivered through its webapp and mobile applications.

> "Service" means the software-as-a-service offering made available by TITLEMAN Inc, including through its webapp, mobile applications
> — https://titleman.ai/legal/platform-agreement · Platform Agreement, section 13.27 Defined Terms

**Public API** · Partly — The only documented programmatic surface is the MCP server, with one HTTP endpoint, OAuth 2.1 and a published tool list; no general REST API is published, and Zapier, webhooks and SFTP are set up on request instead.

> One endpoint, https://titleman.ai/_mcp/titleman, over Streamable HTTP. Every message is its own POST and is answered with JSON; the server keeps no session between requests.
> — https://titleman.ai/docs/claude-connector · Connector documentation, Protocol section

**MCP server** · Yes — A remote MCP server is published with a server URL, OAuth 2.1 with PKCE, fourteen read-only tools and eight paid write tools, and the same connector is documented for Claude, ChatGPT and Copilot Studio.

> Paste the server URL: https://titleman.ai/_mcp/titleman
> — https://titleman.ai/docs/claude-connector · Connector documentation, 'Connect it in Claude or ChatGPT', step 2

**Webhooks** · Partly — A webhooks integration page exists but describes the pairing in the conditional and says every integration is set up on request by the Titleman team, so there is no self-serve event subscription a buyer can switch on.

> On request: we connect Titleman to Webhooks for your team. ... Every integration is set up on request.
> — https://titleman.ai/integrations/webhooks · Webhooks integration page, intro line and 'Request this integration'

## What it costs to start

**Free tier** · No — There is no free plan and the vendor states outright that there is no trial: access opens only after pilot terms are agreed in writing and the pilot fee is settled, and usage runs on prepaid credits.

> No trial. A pilot on agreed numbers. ... You reply yes to that page and the pilot fee is settled. Access for your team opens then, and not before.
> — https://titleman.ai/demo · Demo page, 'How a Titleman pilot works', heading and step 03

**Self-serve signup** · No — Every call to action is a demo request, accounts are opened for firms by the vendor, and logins are created from names the client supplies, so a buyer cannot start without talking to the team.

> accounts are opened for firms — no account yet? request a demo
> — https://titleman.ai/docs/claude-connector · Connector documentation, 'Connect it in Claude or ChatGPT', step 3

**Priced per seat** · Partly — The vendor's own pages disagree: teams get a login for each person and the service level terms compute credits off a Per Seat Cost, while the homepage says the work is priced by the hour it replaces with no seats.

> Service Level Credit based on a percentage of Your Per Seat Cost (as set forth in Your currently operative Order Form) multiplied by the number of Seats
> — https://titleman.ai/legal/support-and-service-level-terms · Support and Service Level Terms, section 2.3

**Cancel any time** · Partly — Either party can terminate on notice, but it only takes effect when open order forms end, the pilot leads into an annual agreement, and prepaid credits are non-refundable and expire 99 days after purchase.

> All purchases of credits are payable in advance and are non-refundable, except as expressly required by applicable law.
> — https://titleman.ai/legal/platform-agreement · Platform Agreement, section 7.1 Fees and Payments

## What happens to your data

**No training on your data** · Yes — The Platform Agreement commits the vendor and its subprocessors to no model training on customer content and no retention or human review of it, and the security page repeats the commitment for rent rolls and data rooms.

> TITLEMAN Inc will not train any AI models using Your Content or Customer Data. Subprocessors will not train any AI models using Your Content or Customer Data.
> — https://titleman.ai/legal/platform-agreement · Platform Agreement, section 12.8 No Training

**Retention you control** · Partly — Deletion is contractual rather than a setting: data is returned or deleted on request or within 30 days of the service ending, and customer data is kept up to 90 days, but no page offers a customer-adjustable retention window.

> On your request, or within 30 days after the service ends, we return or delete your data and direct every subprocessor to do the same.
> — https://titleman.ai/security · Security page, 'What the contracts say', Deletion card

**Data stays in your region** · Partly — Storage regions are agreed in the contract and a different region may be requested, but it is subject to feasibility rather than offered as a choice, and the privacy policy names German data centers as where data is hosted.

> Requests to store Customer Data and Content in alternative geographic regions may be considered, subject to technical feasibility and compliance with applicable laws and regulations.
> — https://titleman.ai/legal/security-addendum · Security Addendum, section 2.2 Data Storage and Processing

**Self-hosted** · Partly — Corporate engagements can run inside the institution against its own policy set and systems of record, which is offered as a private deployment on enterprise terms rather than software a buyer installs and operates itself.

> Private deployment available. Corporate engagements run inside the institution, against your policy set and systems of record.
> — https://titleman.ai/security · Security page, Deployment card

**Open source** · No — Nothing is published under an open license and the terms forbid reverse engineering the underlying model and systems, while the open source clause only promises the Service will not infect customer software with a copyleft license.

> attempt to reverse engineer or attempt to discover the source code or engineering of the underlying model and systems of the Service
> — https://titleman.ai/legal/platform-agreement · Platform Agreement, section 4.2.3 Use Restrictions

**Take your data out** · Yes — Work leaves as ordinary files rather than locked records: decks as editable PowerPoint and PDF, models as live-formula Excel, comps and lists as CSV, and contract terms require return or deletion of customer data.

> OMs, listing pitches, IC memos and lender packages as editable PowerPoint, PDF and a share link, underwriting already inside.
> — https://titleman.ai/facts · Facts page, Products list, CRE decks row

## Whether a company can buy it

**Audited security** · No — No audit report is published and the privacy policy describes SOC 2 Type II and ISO 27001 only as programs underway, with the Security Addendum offering information about practices on request instead of a certification.

> Real-time monitoring and centralized logging; SOC 2 Type II and ISO 27001 certification programs are underway.
> — https://titleman.ai/legal/privacy-policy · Privacy Policy, section 10 Information Security and Accuracy, Security Monitoring bullet

**Audit log** · Yes — The vendor records who ran what, on which sources and with which template, and the connector additionally logs which tool ran, when, how long it took and whether it succeeded for each call.

> Audit. A record of every run. Who ran what, on which sources, with which template.
> — https://titleman.ai/security · Security page, Audit card in the four-card row

**Uptime promise** · Yes — A written 99.5% monthly availability commitment with a defined uptime calculation and tiered service credits of 10, 25 or 50% of per-seat cost, excluding unpaid, evaluation and preview use.

> TITLEMAN Inc will use commercially reasonable steps to make the Services available to You at least 99.5% of each calendar month (the "Availability Commitment").
> — https://titleman.ai/legal/support-and-service-level-terms · Support and Service Level Terms, section 2.1

## A brief goes in, finished CRE files come out

Titleman describes itself as [an AI engine for commercial real estate analysis](https://titleman.ai/facts): it reads a deal's own documents, enriches them from county, tax, zoning, listing and ownership records, and returns underwriting, value opinions, comps, feasibility studies, decks and diligence findings with the source and date of every figure.

The work is organized as six modules over one shared property record: decision support and underwriting, analytics, CRE decks, sourcing, legal and diligence, and CRM. The vendor's argument for that shape is reconciliation rather than breadth: "the comp set that priced the asset is the one printed in the OM and cited in the memo". Deliverables leave as PowerPoint, PDF, Excel with live formulas and CSV, and the [product page](https://titleman.ai/platform) puts the limit of the automation plainly: output is a draft with confidence "clamped to the weakest input", and a named professional on the client's team signs it.

## Pricing is a scoped engagement, not a rate card

There is no published price. The [plans page](https://titleman.ai/plans-for-teams) says so in terms: "There is no list price on this page: the engagement is scoped with you, and the price is something we discuss when you ask." Nothing else on the site, including the FAQ-style pages and the facts page, names a figure for the product itself.

Underneath the engagement, metering is credit-based. The [Platform Agreement](https://titleman.ai/legal/platform-agreement) states that access is "provided on a prepaid, credit-based basis", that each credit entitles the customer to one permitted use, that credit prices are displayed at purchase, and that unused credits "expire ninety-nine (99) days after the date they are added" and are non-refundable. Running a report over the MCP connector spends the same credits as the web app.

## There is no trial, and the way in is a paid pilot

The [demo page](https://titleman.ai/demo) heads its process section "No trial. A pilot on agreed numbers." The sequence is a call, a one-page set of pilot terms covering success numbers, deals, people, dates, price and a decision date, then: "You reply yes to that page and the pilot fee is settled. Access for your team opens then, and not before." The pilot runs 14 days at most, and its fee counts toward the annual agreement if the numbers are met. One benchmark run on a deal the firm already closed is offered free of charge.

## The MCP connector is the real programmatic surface

Titleman publishes a remote MCP server at `https://titleman.ai/_mcp/titleman` and [documents it properly](https://titleman.ai/docs/claude-connector): OAuth 2.1 with PKCE, dynamic client registration, discovery documents, Streamable HTTP with no session, batches of up to ten JSON-RPC messages, and 30-day access tokens. The same connector is documented for Claude, ChatGPT and Microsoft Copilot Studio, and ChatGPT deep research reads reports through `search` and `fetch`.

Fourteen read-only tools cost nothing and start nothing, including comparables on file, a zoning envelope for a parcel, and questions against the account's own deal documents with each passage cited to its page. Eight write tools start real runs and spend credits: BOV, comps, underwriting, highest-and-best-use, buyer list, due diligence, property search and a saved workflow. `run_workflow` is the one that asks first, returning an estimated charge and starting only when called again with `confirm: true` inside a `max_usd` of at most $50.

The vendor's own [guide to the connector](https://titleman.ai/resources/guides/connect-to-claude) is unusually candid about the gaps: "Value Opinion, Scout, Feasibility, Excels and Legal are not exposed over MCP", the connector covers six of the eleven products, and Buyer List has had one client run in thirty days. No general REST API is published; Zapier and Slack are described as not open yet.

## Security is contractual, and the certificates are not there yet

The [security page](https://titleman.ai/security) makes four commitments and points each at a document: firm isolation with nothing pooled across customers, no training on customer content, a record of who ran what on which sources, and private deployment for corporate engagements. The no-training promise is real contract text, not a marketing line: section 12.8 of the Platform Agreement binds the vendor and its subprocessors, and adds that subprocessors "will not retain or log for human review" customer content. The DPA commits to return or deletion on request or within 30 days of the service ending, and new subprocessors are published 30 days before they start.

What is missing is an audit. The [privacy policy](https://titleman.ai/legal/privacy-policy) says "SOC 2 Type II and ISO 27001 certification programs are underway", and the [Security Addendum](https://titleman.ai/legal/security-addendum) is written in the language of reasonable measures, with security information available "upon reasonable request" and testing "at TITLEMAN Inc's discretion". No SOC 2 report, no ISO certificate and no penetration test summary is published.

Residency also needs a question in the sales call. The Security Addendum says data sits in "data centers located in agreed geographic regions" and that other regions "may be considered, subject to technical feasibility", while the privacy policy states data is hosted in ISO 27001-certified data centers in Germany. Those two statements are not the same answer, and only one of them is in an addendum a buyer signs. Neither SSO nor SCIM is mentioned anywhere.

## The claims are the vendor's own measurements

Titleman quantifies itself more than most: up to 95% of a professional's routine work moved to AI, 337 built-in workflows, 2,100+ sources with 1,882 of them having returned data, 1,735 counties across 43 states, and median run times per product measured on production, from about 16 minutes for rent analysis to about 58 minutes for underwriting with its Excel model. The [facts page](https://titleman.ai/facts) dates each count and names the measurement window.

Every one of those numbers is measured by the vendor on its own runs and client engagements, and the footnote says so: "Figures vary by firm and job type." Funding and revenue are listed as not disclosed. The company is TITLEMAN Inc, a Delaware corporation, with Alex Galt named as founder on the [company page](https://titleman.ai/company) and no other named executives or investors published.

## Where it fits, and where it does not

It fits a CRE desk that already produces BOVs, OMs, IC memos and underwriting models on U.S. assets and wants the assembly work done before a licensed person signs. The property passport, the entity resolution through LLC chains and the cited comp set are aimed at work that has to survive an investment committee, and the lender pages extend the same engine to a loan request package.

It does not fit a buyer who wants to try software this afternoon: there is no free plan, no trial and no self-serve signup, and the first commitment is a priced pilot. It does not fit non-U.S. portfolios outside the four countries where registers exist. It does not fit a team that needs an appraisal, legal advice or a title search, which the vendor disclaims on nearly every page: "Titleman is not a broker, appraiser or law firm." Teams that need a published SOC 2 report, SSO or a predictable monthly bill will be waiting on the roadmap or on a negotiation.

## What changed

- 2026-10-20 — Capabilities · Free tier
- 2026-10-17 — Capabilities · Free tier · Price

## Keeping it current

These facts are re-checked against Titleman's own pages and updated when they change. If you represent Titleman, sign in and update the listing from your dashboard at https://agentsindex.ai/dashboard. The operations this site accepts from an agent are listed at https://agentsindex.ai/.well-known/agents.json — that catalog needs no key to read, and each entry states its own auth.