Pilot pre-flight

What it takes to land a 90-day Pilot in two weeks.

Pre-flight checklist for the $15K Pilot rung. Use this to scope a kickoff with us, brief your security and IT teams, and align stakeholders before the architecture call.

Scope reminder

What Pilot is, in one paragraph

The Pilot rung is a production deployment of Bolt and Aeira inside your VPC, on-prem datacenter, or cloud account. Docker compose for single-host scenarios; Helm chart for multi-node Kubernetes. Up to 30 seats. 90-day founder-led engagement plus a weekly 30-minute check-in for the full duration. Flat $15,000, or 5+10 split (5K at kickoff, 10K at day 30). $15K credits against year-one annual if you convert to Founding Customer.

Pilot is not the sidecar deployment. That is the Discovery rung — 30 days, up to 10 seats, $5K, a single binary that runs locally on each user's laptop with no server install. Most customers do Discovery or Pilot first based on team size and whether they have IT to support a deployment, not both.

Production Docker or Helm Up to 30 seats 90 days $15K flat or 5+10 split Founder-led

Phase by phase

The 12 weeks, week by week

Week -2 to -1: Architecture + scoping

Owner: Sparcle founder + your Innovation/AI lead
  • 30-minute architecture call. Confirm deployment topology: Docker single-host (docker-compose.ha.yml) or Helm multi-node on your Kubernetes.
  • Confirm cloud target (AWS / Azure / GCP), on-prem datacenter, or VPC.
  • Identify the 20 to 30 pilot users (named list, SSO accounts).
  • Identify the corpora to ingest (DMS folders, mailbox scopes, file shares, code repos, ticketing, CRM). Cap default 250GB.
  • Choose LLM endpoint: Azure OpenAI, AWS Bedrock, Google Vertex, Anthropic, on-prem Llama, or other OpenAI-compatible. Share endpoint URL + key handling preference.
  • Choose 2 to 5 MCP connectors to wire from the standard catalog (Microsoft 365, Google Workspace, Slack, GitHub, Jira, Salesforce, Confluence, ServiceNow, etc.).
  • Pre-filled SIG Lite + draft DPA + draft MSA delivered for legal review.

Week 0: Kickoff + paperwork

Owner: You + Sparcle founder
  • Counter-sign MSA + DPA. Optional: 5+10 payment plan (5K at kickoff, 10K at day 30).
  • Confirm pilot acceptance criteria in a one-page outcomes note (we provide template).
  • Schedule recurring weekly 30-minute check-in for the next 12 weeks.
  • Provision the deployment target (single VM, AKS/EKS/GKE cluster, or on-prem host) with the requirements doc you receive at architecture call.
  • Stand up SSO connection (Entra / Google / Okta / OneLogin / Auth0 / any OIDC).

Week 1 to 2: Production deployment

Owner: Sparcle (founder-led)
  • Bolt + Aeira services deployed via Docker compose or Helm chart inside your infrastructure. Same product image used at Founding Customer scale.
  • Per-tenant CMK (envelope encryption keys) provisioned. Crypto-shred enabled.
  • Sealed Merkle audit chain initialized. Audit verifier endpoint exposed to your security team.
  • SSO + SCIM 2.0 wired to your IdP. First 5 users provisioned end-to-end as smoke test.
  • LLM endpoint connected and round-tripped.
  • MCP connectors authenticated against your identity stack.
  • Aeira ingest sagas configured for the chosen corpora. First ingest run completes.

Week 3 to 10: Real usage, weekly check-ins

Owner: You (your users) + Sparcle (founder support)
  • All pilot users provisioned. Onboarding walk-through delivered (60 minutes, recorded).
  • Weekly 30-minute check-in to triage friction, add connectors on request, tune Aeira ranking, surface adoption metrics.
  • Dedicated Slack channel between your pilot team and the founding team for the duration.
  • Security team validates: audit chain on every query, ACL enforcement across personas, crypto-shred test on at least one user's data.
  • Optional: CISO-led architecture review using the threat model document we share at kickoff.
  • Mid-pilot health check at week 6: continue, pivot scope, or end early.

Week 11 to 12: Outcomes + decision

Owner: You + Sparcle
  • Draft one-page outcomes note jointly. Three to five workflows your users say Bolt has changed for them.
  • Audit chain export + verifier walk-through for your compliance team.
  • TCO walkthrough for finance: per-seat cost at Founding Customer terms vs. Pilot list price.
  • Decision: convert to Founding Customer (25% off list, 24-month lock, $15K credits year one), continue at standard pricing, expand the scope, or end with no further obligation.
  • If converting: counter-sign annual MSA + revised DPA. No reinstall required — same deployment continues under the new contract.

Division of labor

What you bring, what we bring

You bring

  • Deployment target — one Linux host (Docker compose path) or a Kubernetes cluster (Helm path). 4 vCPU / 16GB RAM is the comfortable floor for 30 seats.
  • LLM endpoint — Azure OpenAI, AWS Bedrock, Google Vertex, Anthropic API, OpenAI API, on-prem Llama / vLLM, or any OpenAI-compatible endpoint. Bring your own API key.
  • IdP — Microsoft Entra, Google Workspace, Okta, OneLogin, Auth0, or any OIDC provider. We need OAuth Client ID + Secret, and a service account for SCIM if you want provisioning.
  • Pilot user list — up to 30 named users with SSO accounts. They will use Bolt as their daily AI tool for the duration.
  • Corpora list — the DMS folders, mailbox scopes, drive folders, code repos, ticketing, CRM scopes you authorize for indexing. Up to 250GB default; larger by scope letter.
  • Security reviewer — one named contact who will run the architecture review, sign off on the DPA, and validate the audit chain.

Sparcle brings

  • Production deployment — Docker compose or Helm chart, founder-led install.
  • Founder onboarding — kickoff call, weekly 30-min check-ins for 12 weeks, dedicated Slack channel.
  • Engagement scope — 90 days of founder-led support for up to 30 users.
  • SIG Lite + threat model + DPA + MSA drafts for your security and legal review.
  • 2 to 5 MCP connectors wired from our standard catalog.
  • Architecture diagram + audit-chain primer for your CISO conversation.

Acceptance criteria

What “successful Pilot” means at day 90

We sign off on a one-page outcomes note together at week 12. Default criteria below; adjust by agreement in the kickoff call:

  • Adoption: 70%+ of pilot users log in at least once per week by week 8.
  • Workflow impact: each user can name 3 to 5 workflows Bolt has changed for them at day 90.
  • Governance: audit chain verifier passes on every query in the pilot window. Crypto-shred test succeeds on at least one user.
  • Performance: p95 query latency stays within your environment's expected range (architecture call sets the target).
  • Security sign-off: your CISO or security lead countersigns the threat model + DPA review with no blocking issues.

Out of scope

What the Pilot does not include

  • Air-gapped or classified-network installs — that requires the Federated Aeira tier and a separate architecture engagement.
  • Custom MCP connector development beyond the standard catalog (available as paid Customization).
  • Custom Aeira authority policy or reranker beyond the 3 reference impls (Default, Regulated, Healthcare).
  • FedRAMP / IL4 / IL5 deployment (not authorized today).
  • 24x7 production SLA (that ships at Founding Customer).
  • Multi-region / multi-cluster failover (Founding Customer or later).

Ready? Architecture call is the next step.

One 30-minute call to confirm topology (Docker or Helm), cloud target, LLM endpoint, IdP, pilot users, and connectors. We send the SIG Lite, DPA, and MSA drafts within 48 hours of that call.

Or email bolt@sparcle.app · smaller scope? see Discovery