Telos Private Office · March 2026
10 autonomous agents to replace repetitive operational work across finance, compliance, scheduling, and business development.
Key Distinction
Both use AI. The difference is who drives.
Could a junior-to-mid-level employee do this if you handed them the inputs and told them what output you wanted? If yes, it's an agent.
The Portfolio
Organized by function. Each runs autonomously on schedule or in response to events.
Agent 01
Watches for new transcriptions from Plaud, Wudpecker, or Otter. Automatically parses speaker attribution, extracts key decisions with owners and timelines, pulls out action items with priorities, and formats everything into a structured markdown summary.
Triggered when a new transcript file lands in a watched Drive folder. The agent reads the full transcript, identifies speakers, categorizes discussion topics, and outputs a structured document with Decisions, Action Items (table format), Discussion Summary, Follow-ups, and Parking Lot items. Posts to Slack for team review.
Alyssa manually processes meeting transcripts: reading through 30-60 minute recordings, hand-extracting action items, formatting notes, and distributing to the team. This takes 20-40 minutes per meeting, sometimes longer for complex sessions.
At 4-6 meetings/week, 20-40 min each. Agent reduces to a 2-minute review of the formatted output.
Agent 02
Takes a meeting request (who needs to meet, about what, by when) and produces a polished RAP (Rapid Action Plan) block: 3 proposed time slots pulled from real calendar availability, attendee list, purpose statement, pre-read materials, and response deadline.
Triggered via Slack command or as a downstream agent after Meeting Notes Processor identifies follow-up meetings. Pulls calendar availability via Google Calendar, finds 3 genuinely different time windows, formats the RAP block, and posts to Slack. On approval, creates the calendar event and sends the invite via Gmail.
Scheduling executive meetings requires checking multiple calendars, proposing times, formatting the request professionally, and following up. The back-and-forth typically takes 15-30 minutes per meeting and often spans multiple days.
At 5-8 scheduling requests/week, 15-30 min each. Agent handles the full cycle in under a minute, including calendar lookups.
Agent 03
After Outrider's prospect-research agent completes a research brief, this agent drafts personalized outreach emails. Handles cold outreach, warm intros, follow-ups, deal introductions, and re-engagement. Each email is under 200 words with a specific value prop and clear CTA.
Runs automatically after prospect research completes. Reads the prospect brief, identifies the best email type, drafts 1-2 variants with personalized hooks tied to the prospect's situation. Posts drafts to Slack for Davy's review. On approval, queues via Gmail MCP.
Davy manually writes outreach emails after reviewing research, spending 15-25 minutes per prospect crafting personalized messaging. The gap between research completion and outreach often means lost momentum.
At 8-12 prospects/week, 15-25 min each. Agent drafts immediately after research completes. Davy reviews and sends in 2-3 minutes.
Agent 04
Generates a branded single-page portfolio summary for new clients. Includes company overview (2-3 sentences), key metrics (3-5 numbers), services provided, key contacts, and next steps. Output follows one of three layout options: left-right split, top-bottom, or card-based.
Triggered when a new client folder is created in Drive or via Slack command. Agent pulls available data from the onboarding docs, applies the appropriate portfolio brand template, generates the HTML one-pager, and uploads it to Drive. Posts a preview to Slack for Greg's approval.
Greg creates client one-pagers manually, gathering data from multiple sources, formatting to brand guidelines, and ensuring every word earns its space. Takes 45-90 minutes per client, and often gets deprioritized.
At 2-4 new clients/month, 45-90 min each. Agent produces a review-ready draft. Greg refines in 10 minutes.
Agent 05
Monitors Delaware annual filing deadlines across all Telos entities. Calculates franchise tax using both methods (Authorized Shares and Assumed Par Value), recommends the lower amount, verifies registration status, and alerts before deadlines. Tracks annual report requirements and reinstatement procedures if needed.
Runs on a monthly schedule with increased frequency in April-May (ahead of June 1 deadline). Reads entity data from the project context, calculates tax obligations per entity, checks status against last year's filing records, and posts a compliance summary to Slack. Flags any entities at risk of penalties.
Katherine manually tracks filing dates across 10+ entities, calculates franchise tax using both methods to find the lower amount, and cross-references registration status. A missed deadline means a $200 penalty plus 1.5% monthly interest.
Concentrated around filing season. Agent eliminates the manual calculation and tracking entirely. Katherine reviews and files.
Agent 06
Generates consolidated monthly budgets across all Telos entities. Pulls actuals from QBO, projects forward based on historical trends, handles intercompany transaction eliminations, and produces variance analysis comparing actuals to budget. Full P&L structure from Revenue through Net Income.
Triggered at month-end. Connects to QuickBooks to pull actual figures per entity, applies budget assumptions from the project context, eliminates intercompany transactions, consolidates, and generates the multi-tab report. Uploads to Drive and posts a variance summary to Slack highlighting any line items >10% off budget.
Katherine pulls data from QBO for each entity, manually consolidates, eliminates intercompany transactions, and builds the variance report. This is a full-day process every month-end, highly error-prone at the intercompany elimination step.
From a full day of manual work to a 30-minute review of the agent's output. Intercompany eliminations are automated and consistent.
Agent 07
Tracks sales tax obligations across all states where Telos entities have nexus. Performs nexus analysis (physical, economic, click-through, affiliate), verifies registration status per state, reviews filing history for discrepancies between GL and filed returns, and builds a 12-month filing calendar with all deadlines.
Runs monthly, pulling sales data from QBO to check against state thresholds (most states $100K, CA $600K, WA $1K). Cross-references filed returns against GL totals to catch discrepancies. Posts a compliance dashboard to Slack showing upcoming deadlines, registration status, and any flagged issues.
Multi-state sales tax compliance requires tracking different thresholds, filing frequencies (monthly/quarterly/annual), and registration status. Katherine manually cross-references sales data against each state's rules. A missed filing means penalties and interest.
Monthly compliance review drops from a half-day project to a 15-minute review of the agent's flagged items.
Agent 08
Maps a new client's QuickBooks Online chart of accounts to the universal Telos template (1000-7999 ranges). Analyzes each account's type, balance, and usage patterns. Assigns confidence levels (High/Medium/Low) to each mapping. Generates a formal mapping table with implementation notes.
Triggered when a new client's QBO access is granted. Agent pulls the full chart of accounts via QuickBooks, analyzes each account against Telos template ranges (Assets 1000-1999, Liabilities 2000-2999, etc.), generates the mapping with confidence scores, and posts to Slack for Sarah's review before implementation.
Sarah manually reviews each client's chart of accounts, matches them one by one to the Telos template, documents the mapping, and implements changes in QBO. A single client onboarding takes 2-4 hours of focused mapping work.
Agent produces the full mapping in minutes. Sarah reviews the Low-confidence mappings (typically 10-15% of accounts) and approves.
Agent 09
Validates the newly mapped chart of accounts against best practices. Checks numbering conventions, account name consistency, type classifications, identifies redundant or dormant accounts, and flags consolidation opportunities. Produces a validation report with Critical/Major/Minor severity levels.
Runs automatically after account mapping completes. Reads the mapping output, runs validation checks (non-standard ranges, gaps, duplicates, ambiguous names, misclassified accounts, too-granular breakdowns), and generates an action plan. Posts findings to Slack with a prioritized fix list.
After mapping, Sarah manually audits the structure for consistency issues. This is tedious, detail-oriented work that often gets abbreviated due to time pressure, leading to reporting inconsistencies discovered months later.
Validation runs automatically after every mapping. Catches issues that manual review misses. Sarah focuses only on the flagged items.
Agent 10
Checks 1099-NEC compliance across all vendors. Identifies vendors approaching or exceeding the $600 threshold, verifies W-9 status, confirms entity type exemptions (C-Corps exempt), and produces a filing-ready compliance checklist. Catches missing W-9s before the January 31 deadline.
Runs quarterly with an intensive year-end pass. Pulls vendor payment data from QBO, cross-references against W-9 records, applies exemption rules (C-Corps, certain LLCs), flags vendors within 80% of the threshold as "approaching," and generates the complete 1099-NEC filing list with action items for missing documentation.
Year-end 1099 prep is a scramble. Sarah manually pulls vendor lists, checks payment totals, chases down missing W-9s, and verifies entity types. The January 31 deadline creates a compressed, stressful window. Quarterly checks would help but rarely happen due to time constraints.
Year-end prep drops from 2-3 full days to a half-day review. Quarterly checks happen automatically, catching issues early instead of in January.
Recommendation
Two agents that deliver value fastest with the simplest setup.
Week 1: Configure Google Drive + Calendar integrations, build agent prompt templates, set up Slack channels | Week 2: Deploy Meeting Notes Processor, run in shadow mode alongside manual process | Week 3: Deploy RAP Scheduler chained to Meeting Notes, begin live operation | Week 4: Review, tune, plan next agent batch
Next Steps
10 agents. 30+ hours saved per month. Start with two, prove the pattern, scale from there.
Approve the Meeting Notes + RAP Scheduler as the first two agents. Kairos will have them running within two weeks.
Kairos Garage · kairosgarage.com