Training

Three tracks, each ending with something your team can use.

A general AI program for lending teams, and two focused workshops on the infrastructure deals now reaching credit desks: building the model behind them, and underwriting them.

Book a discovery call See the tracks
Drawing: the track plan Three training tracks drawn as lanes. AI for Lending Teams climbs six stages from prompt to build. Technical Financial Models adds six tabs to one model, from capital structure to the committee summary. Underwriting Datacenter Deals works through the four tests: installed, powered, paid and recovered. All three lanes end in the same box: something your team built and keeps. Summit Logic Track plan Dwg SL-0011 Sheet 1 of 1 Rev A Not to scale 1 2 3 AI for Lending Teams Financial Models Datacenter Deals Keynote and modules Half day By request Prompt Context Automate Agent Orchestrate Build 1 Capital structure and debt sizing 2 Depreciation by asset class 3 Deposits and draws 4 Loan balance against resale 5 Stress tests 6 The committee summary Installed Powered Paid Recovered Site, racks Power, timing Customer, term Residual value Something built A workflow, a model, a checklist. Yours to keep.
the tracks

Pick the room

  1. AI for Lending Teams

    Lenders, brokers, credit analysts and loan operations teams.

    A keynote and eight core modules built on lending documents: intake, extraction, spreading and package assembly, with the credit decision kept human.

    Format
    Keynote and eight modules
    Availability
    Book now
    Jump to the track

  2. Technical Financial Models

    Credit analysts, originators, CFOs and FP&A teams.

    Build and read a lender-ready infrastructure model: debt sizing, split depreciation, deposits and draws, the recovery curve and the stress tests.

    Format
    Half day, or two 2-hour sessions
    Availability
    Book now
    Jump to the track

  3. Underwriting Datacenter Deals

    Equipment finance lenders, brokers and credit teams.

    Our four-test framework, installed, powered, paid and recovered, and a prequalification checklist for the next datacenter or GPU deal.

    Format
    Two hours, or a half day for credit teams
    Availability
    By request
    Jump to the track

AI for Lending Teams

For the people who receive the documents, key them, chase what is missing and assemble the package, and for the managers who decide what to automate next. A keynote that shows all six stages in ninety minutes, eight core modules that climb them one stage at a time on your own lending documents. No technical background is assumed.

Book a discovery call See the modules
Drawing: the capability ladder Six stages drawn as a stair, rising from prompt to context, automate, agent, orchestrate and build. The eight core modules are numbered on the stage each one teaches. Summit Logic Capability ladder Dwg SL-0012 Sheet 1 of 1 Rev A Not to scale Capability Prompt Brief it like an analyst Context Load the workspace first Automate Runs the same every time Agent Multi-step, with gates Orchestrate Supervisor and workers Build A tool of your own 1 2 3 4 5 6 8 7 Core modules
format

A keynote, then eight modules

Keynote, 90 minutes Touches all six stages with a working demo on each. Sets the vocabulary and shows what the eight modules build. A 60-minute cut keeps both live builds.
Core, eight modules Ninety minutes each, in person or on Zoom, scheduled to fit your team. Teach, live demo, hands-on breakout, report back, one assignment.

What you leave with

  • A working intake, extraction or packaging workflow on your own documents
  • A written automation map of your loan workflow
  • A ninety-day roadmap your team can run
  • The rules for what stays human, written down

Eight modules, one stage at a time

Each module names the stage it teaches, from prompt to build, so a wide scope never feels like a catalog.

  1. 01

    Prompt: AI fluency and the briefing discipline

    What models do and where they fail, and what never gets pasted. Leaves with three prompts you will use right away.

  2. 02

    Context: context engineering

    Why the same prompt behaves differently in a loaded workspace. Leaves with a live project that makes later modules sound like you.

  3. 03

    Context: borrower communication at scale

    Request letters, status updates and missing-item chases from one brief, in your voice. Leaves with a template set your team will actually send.

  4. 04

    Automate: the automation layer

    Intake routing, missing-item follow-up, broker and borrower status updates, and pipeline hygiene. Leaves with a written automation map of your own loan workflow.

  5. 05

    Agent: your first agent

    What an agent is versus a chat; setup, plans, permissions and approval gates. Leaves with one real multi-step task, such as assembling a document checklist, completed by an agent instead of by you.

  6. 06

    Agent: extending the agent

    Connectors, skills, scheduled work and subagents. Leaves with a scheduled or repeatable job running in your own environment.

  7. 07

    Build: from description to working software

    Iteration without breakage, and hosting realities. Leaves with a functioning tool you built, shareable by link.

  8. 08

    Orchestrate: orchestration and the roadmap

    Who on your team should reach the build stage, supervisor and worker patterns, and the audit. Leaves with a prioritized ninety-day roadmap you can actually run.

Technical Financial Models

Build and read a lender-ready infrastructure model. For credit analysts and equipment finance originators seeing datacenter and compute deals, and for CFOs and FP&A teams evaluating an infrastructure investment of their own. Comfort with a spreadsheet is enough; no engineering background is assumed.

Book a model workshop
Drawing: value against loan Over five years of asset life, facility equipment loses value slowly and GPU servers lose value quickly, while the loan balance amortizes in a straight line. Facility value stays above the balance; GPU value falls below it, which is why the model depreciates the two classes separately. Summit Logic Value against loan Dwg SL-0003 Sheet 1 of 1 Rev A Illustrative Value Year 1 Year 2 Year 3 Year 4 Year 5 Asset life Cost Facility equipment GPU servers Loan balance GPU value below balance
format

A half day, or two sessions of two hours

Length A half-day workshop, or two 2-hour sessions a week apart.
Hands on Every participant works in the template, not just watches it.
Group size Sized for a credit team or a finance department, so every model gets looked at.
Materials The template, a worked example deal and the stress-test set, kept by your team afterwards.

What you leave with

  • A working model template you can use on the next deal
  • The ability to build a lender-ready infrastructure model, and to read one critically
  • A stress-test set: customer loss, power delay and residual shock
  • A one-page summary format a credit committee can read

Six modules, each built into the template

Every module adds a tab to the same model, so the template is complete when the workshop is.

  1. 01

    Capital structure and debt sizing

    Sizing the debt to a coverage target, and what the equity gap really is once deposits and draws are timed.

  2. 02

    Depreciation by asset class

    Why switchgear and GPU servers do not share a curve, and what happens to a model that pretends they do.

  3. 03

    Procurement, deposits and draws

    Long-lead equipment, deposit schedules and milestone-based draws, and the cash they need before anything earns.

  4. 04

    Loan balance against resale value

    Building the recovery curve: what the lender gets back at each point in the term, by asset class.

  5. 05

    Stress testing

    Customer loss, power delay and a residual-value shock, each run against coverage and the recovery curve.

  6. 06

    Presenting to a credit committee

    The one-page summary, the questions a committee will ask, and how to answer them from the model.

Underwriting Datacenter Deals

For equipment finance lenders, brokers and credit teams whose files now include GPU servers, switchgear and liquid cooling. Our four-test framework, worked through on a live example, taught by someone who has designed and run the facilities. The first module covers what a facility actually needs, in plain English.

Request a workshop
Assembly drawing of a compute deal file Three public signals (filings, power and permit records, hiring and funding news) feed a qualification score of fit and intent. Qualified deals pass four tests: installed, powered, paid and recovered. The technical financial model plots facility equipment and GPU servers on separate depreciation curves against the loan balance. The file goes to the lender as a one-page summary with the evidence for each test, signed off by a credit committee. Summit Logic Compute deal file Dwg SL-0002 Sheet 1 of 1 Rev A Not to scale 1 Signals Public filings Power and permits Hiring and funding Watched daily 2 Qualification Fit Intent Fit x intent 3 Four tests Installed Powered Paid Recovered Site, racks Power, timing Customer, term Residual value 4 Financial model Value against loan Facility GPU Loan 5 Lender Test Finding Evidence Installed Colocation hall, 2 MW Lease, section 4 Powered Utility service confirmed Power agreement p.2 Paid Five-year customer contract MSA, schedule B Recovered Resale above loan balance Residual curve, sheet 4 Committee
format

Two hours for broker and lender groups, a half day for credit teams

Short version A two-hour workshop for broker and lender groups.
Long version A half day for credit teams, with more time on the live example.
Availability Offered by request while the format is piloted with lender groups. Ask and we will schedule one.

What you leave with

  • Our four-test framework: installed, powered, paid, recovered
  • A prequalification checklist to use on the next deal
  • The red flags and knockouts that end a deal early
  • A worked example, start to finish

Six modules, from the facility to the knockouts

Each test is taught the way a credit team will use it: the question, the document that answers it, and what a bad answer looks like.

  1. 01

    What a datacenter actually needs

    Space, power, cooling and density, in plain English, and why each one changes the credit.

  2. 02

    Test 1, Installed

    The site, colocation contracts, landlord waivers and rack density: is there a real place to put the equipment?

  3. 03

    Test 2, Powered

    Power contracts, cost per kilowatt-hour, moratoriums and long-lead electrical gear: will it have power, when, and at what cost?

  4. 04

    Test 3, Paid

    Customer credit, contract length against the loan term, and concentration: who pays, and for how long?

  5. 05

    Test 4, Recovered

    GPU residuals, the secondary market, removal costs and mixed asset lives: what does the lender get back?

  6. 06

    Red flags and knockouts

    The answers that end a deal early, then a live example worked through all four tests.

What training does not do

No track is taught as a replacement for credit judgment. The workshops teach how to build and read a model and how to organize the evidence on a datacenter deal; they do not make a credit decision, value a specific asset, or give investment or credit advice on a specific deal. Every room names what stays human, and every tool is checked the week before it is taught, because this part of the stack changes faster than any curriculum.

Start with a training day

Tell us who will be in the room and the work on their desks. We will propose the track, the format and a worked example that fits.

Book a discovery call