Choosing the Right AI Model: Claude vs GPT vs Gemini vs Copilot

​Meet Dana. She runs operations at a 40-person multifamily owner-operator in Atlanta. Her firm pays for Claude, ChatGPT, Gemini, and Copilot, roughly $200 a month in seats that nobody can justify individually. Her acquisitions lead swears by one, her marketing hire swears by another, and Dana has to decide what the firm standardizes on before renewal.

​Every provider claims to be the best choice, and each one is right about something. They are not competing to build the same product. Anthropic, OpenAI, Google, and Microsoft are four companies with different businesses, different customers, and different reasons to exist.

​Over two hours we work through both halves of the decision. First, the strategic layer: how each provider is built, who they are building for, what that means for pricing, data handling, integrations, and where each company is likely to invest next. Then the practical layer: a set of real estate workflows run head to head across all four, so the differences stop being abstract. Document abstraction, financial modeling, market research, long-context deal review, image and site plan work, drafting in a firm's voice. Same task, four models, visible results.

​We also spend real time on the part most comparisons skip: how to push a model to the edge of what it can do. Reasoning effort, context management, connected tools, structured instructions, and the specific habits that separate a $20 subscription used badly from the same subscription used well.

​This is not a session that ends with one winner. It ends with a decision framework Dana can defend at a partner meeting, including the cases where the answer is not the tool Thesis Driven teaches most often.

​You'll learn how to

• ​Read the business model behind each provider
  Understand who Anthropic, OpenAI, Google, and Microsoft are actually selling to, how each one makes money, and why that shapes everything from pricing tiers to data policy to which capabilities ship first.

• ​Match the workflow to the model
  Build a task-to-model map for the work a real estate firm actually does: abstraction, modeling, research, drafting, visual review, and bulk document work.

• ​Push each tool past generic prompting
  Use reasoning controls, context management, and structured instructions to get materially better output from the subscription you already pay for, before concluding you need a different one.

• ​Judge a comparison honestly
  Read benchmarks and model announcements with appropriate skepticism, and evaluate a new release against your own work rather than someone else's test set.

• ​Decide what your firm standardizes on
  Weigh the real costs of one platform versus several: seat spend, training time, data governance, integration depth, and what happens when the leader changes six months from now.

• ​Place a bet you can revise
  Structure your firm's AI tooling so switching costs stay low, since the ranking will change and the framework needs to survive it.

​What the workshop covers

• ​Meet the operator
  We introduce Dana, her firm, and the decision on her desk, four subscriptions and a renewal date, which is the same position most mid-size firms are in right now.

• ​Four companies, four businesses
  How each provider is built and funded, who their real customer is, and what that predicts about capability, pricing, and roadmap. The strategic context that makes the rest of the session make sense.

• ​Head to head on real work
  Running the same real estate workflows across all four models live, with the outputs side by side. Where the differences are real, where they are marginal, and where they are entirely a function of how the prompt was written.

• ​Pushing to the edge
  The techniques that widen the gap between casual and serious use of any of these tools: reasoning effort, context strategy, connected tools and integrations, and instruction design.

• ​The decision framework
  Turning everything into a defensible answer: which model for which task, what to standardize on, what to keep as a second seat, and how to re-run the decision when the next model ships.

​Who this is for

• ​Intro: Operators, principals, and firm leaders deciding what AI tooling their team runs on, plus the practitioners who want more out of the subscription they already have.

• ​Roles: Principals, COOs, Firm leaders, Acquisitions leads, Asset managers, Investor relations, Marketing leads, Solo GPs

• ​Note: No technical background required. This is a decision-making and technique session, not a coding session.

​Format & access

• ​One live session  2 hours via Zoom, with time for questions throughout.

• ​Live head-to-head demos  We run the workflows across all three models live. Participants are encouraged to follow along in whichever accounts they have.

• ​A recording for everyone  Recordings are sent to all registered attendees, whether or not you attend live.

• ​Slides & materials  Participants receive the slides, the task-to-model framework, and the prompt sets used in the session.

• ​Ongoing access  Your recording, slides, and materials stay in your Thesis Driven account, ready whenever you need them.

​Frequently asked questions

​Do I need accounts on all platforms?
No. Everything is demonstrated live and the comparisons are shown side by side. Having at least one paid account is useful if you want to follow along, but the framework works whether you subscribe to one or all three.

​Is this just going to tell me Claude is best?
No. Thesis Driven teaches Claude most often, and this session is explicit about why, including the workflows where one of the others is the better tool. If the honest answer for a given task is ChatGPT or Gemini, we’ll say so.

​Won't this be out of date in three months?
The rankings will change. The provider business models, the task-matching logic, and the techniques for pushing a model to its edge will not, which is why the session spends more time on the framework than on the current leaderboard.

​Do I need to know how to code?
No. This is about choosing and using tools well, not building with APIs.

​I can't attend live. Will there be a recording?
Yes. Every registrant receives the recording, slides, and materials in their Thesis Driven account.