Deal Screening
During initial screening, an analyst's agent calls the diligence agenda tool with the target's profile and returns a starter agenda the IC memo is built around, recovering 2 to 4 hours per screened deal.
ESEste sitio está disponible en español.
PTEste site está disponível em português.
Use GST's advisory tools (opens in new window) with your AI agents. Supercharge your analysis to accelerate and lower costs of decision-making.
Your analysts already screen targets, portfolio companies and deals with AI agents; connect those AI agents to GST's MCP server to accelerate assessment, benchmarking, and project administration.
During initial screening, an analyst's agent calls the diligence agenda tool with the target's profile and returns a starter agenda the IC memo is built around, recovering 2 to 4 hours per screened deal.
A portfolio operating partner benchmarks total technology spend against stage-adjusted ranges and prices the carrying cost of technical debt, turning "engineering feels slow" into a number the board can act on.
Before bidding on a healthtech target, an agent scopes which frameworks are in play by jurisdiction and effective date, so counsel starts from an exposure list instead of an open question.
A client's internal tooling (Retool, a Slack bot, a custom dashboard) calls the server directly, treating GST's frameworks as a managed API rather than a site somebody has to visit.
The server implements all three MCP server primitives: Tools, Resources, and Prompts. Together they give an agent what a good analyst needs on day one: instruments to run the numbers, the reference material to run them against, and a method for what order to do it in.
The work an associate would otherwise build in a spreadsheet, or wait on an advisor to deliver. Each tool runs a GST model against your deal's parameters and returns figures defensible enough for an IC memo: a scoped diligence agenda, a cost benchmark, a debt estimate, a regulatory exposure list. The model invokes them on its own as the conversation calls for them, so an analyst asks in plain English instead of learning an API. Radar tools are granted separately. Tools in the MCP spec
Turns a target profile into a prioritized diligence agenda with scope and timeline estimates.
Searches GST's engagement record by sector, deal type, stage, and scale, and lists the values each filter accepts.
Queries 123 regulatory frameworks by jurisdiction, category, and effective date.
Scores cloud cost governance maturity across six domains and returns a prioritized improvement list.
Benchmarks total technology cost against stage-adjusted ranges and projects a 36-month trajectory.
Estimates the annual carrying cost of technical debt from team size, salary, and delivery metrics.
Builds a fillable information request list, populates it from evidence already in hand, then ingests the completed file and checks every answer back to its source.
Searches curated PE, M&A, and enterprise-technology intelligence, filtered by category and recency.
The frameworks behind the numbers, readable in full. An agent working from GST's actual VDR taxonomy, diligence checklists, and regulation corpus grounds its answers in a published method rather than in whatever it absorbed during training, which is the difference between a citation you can check and a confident guess. The host application loads these as context, so they are in the conversation before the model needs them. Resources in the MCP spec
The reference guides behind GST's diligence method, in full.
gst://library/<slug>One document per framework, each with its requirements, penalties, and effective dates.
gst://regulations/<jurisdiction>/<framework>FYI carries GST-annotated highlights; Wire carries the full stream, whole or by category.
gst://radar/fyi|wire/<category>Knowing which tool to call is not the hard part. Knowing the order is, and that is what a firm's experience actually buys. Each prompt captures a GST workflow as a sequence, so a first-year associate running a diligence kickoff follows the same progression a GST engagement would. They surface as slash commands in the client, chosen deliberately by a person rather than guessed at by a model. Prompts in the MCP spec
Open a diligence engagement, size up a target quickly, and produce the memos a deal team hands onward.
Issue a request list to a target, populate it from evidence in hand, then sweep what comes back into a cited dossier without rekeying it.
Standing reviews: architecture by layer, regulatory exposure, and the day's radar.
Tiers differ by who is using the server and how broadly, not by which tools you get. Scopes are set per client during provisioning.
free-pilotEvaluation access for a single team. Narrow scopes, radar excluded by default.
Request this tier →enterpriseFirm-wide access across multiple clients, compliance capabilities on request.
Request this tier →| Calls / min | 30 | 60 | 120 |
|---|---|---|---|
| Calls / day | 300 | 2,000 | 10,000 |
| Radar / min | 3 | 5 | 10 |
| Radar / day | 20 | 50 | 150 |
On these numbersThese are tunable, non-contractual capability ceilings, NOT ratified SLA quotas. They are abuse and capacity limits, adjustable per client, and nothing here is a promised allowance.
client_credentials, using a client secret or a registered signing key.One endpoint to configure, once a client record exists. Credentials arrive with provisioning.
Endpointhttps://mcp.globalstrategic.tech/mcp
Four guides take a provisioned client from credential to everyday use: connect the server, drive it well, ingest a completed information request list, and look up any capability's contract.
Connector setup for Claude on the web and Claude Desktop: prerequisites, quick start, and how to verify the connection.
Open guideFirst query, prompt argument forms, pinned resources, and what each failure means.
Open guideIngest a completed information request list and drive nine engines in one turn.
Open guideEvery capability's contract: arguments, a worked example, what comes back, and which workflows use it.
Open referenceTell GST who you are, what you are building, and roughly how much you expect to call. Provisioning is a conversation, not a signup form, which is also how scopes and a tier get set correctly the first time.
REQUEST_MCP_ACCESS()