Newsletters & receipts
Labelled and queued for archiving. You approve the batch in one click, then read them on your own time or never. Either way they do not pollute the inbox.
Task tier · Email triage
Lyra, your AI inbox triage agent.
Lyra watches your inbox in the background. She classifies mail by urgency and topic, drafts replies to common patterns, suggests meeting times, and sends you a daily digest of only the items that actually need you. Most founders reclaim 60–90 minutes per day (typical estimate) after deployment.
ModelFast, low-cost models
Running cost~€0.002–€0.007 per email (€15–€40 a month at 200–300 emails a day)
Connects toGmailOutlookGoogle CalendarCal.comNotionSlack
Autonomy
- Suggests
- Acts with approval
- Acts on its own
Sorts, labels and drafts on its own, and suggests meeting times. You approve every send, booking and archive. How autonomy works
What does Lyra do?
Lyra is an AI email triage agent. She watches your inbox in the background, classifies incoming mail by urgency and topic (intros, vendor questions, scheduling, FYI, noise, urgent), drafts replies to common patterns in your voice, suggests meeting times from your calendar, labels newsletters and receipts, and sends you a daily digest of only the items that actually need your attention. Sending, booking and archiving happen only when you approve. Runs on fast, low-cost models. Task tier, with an eval suite re-run on every change (task agents are more bounded than operational ones). Typical reclaim: 60–90 minutes per day (typical estimate).
theagency47 · Updated October 2026 The spec sheet
How Lyra actually works.
| Job-to-be-done | Reduce email-attention time to a daily digest plus pre-drafted replies, surfacing only what genuinely needs the human |
| Tier | Task (repetitive, well-bounded) |
| Underlying model | Fast, low-cost models, Lyra does many small actions, not few large ones |
| Trigger | Real-time webhook on new email + scheduled daily digest at user-configured time |
| Inputs | Inbox messages, voice document (sample sent emails), calendar availability, contact priority list, exclusion rules |
| Tools available | Gmail/Outlook read, draft, label, archive (on your approval) · Calendar read, booking on your approval (Cal.com or Google Calendar) · Notion (for "save for later" notes) |
| Autonomous decisions | Email classification, labels/tags, archive-vs-surface suggestions, draft language, meeting-time suggestions, daily digest contents. Sending, booking and archiving wait for your approval. |
| Escalation rules | New sender (no prior history) → human · Financial/legal/family keyword → human · Confidence < 0.85 → human · Sentiment angry/urgent → top of digest |
| KPIs measured | Inbox-time reduction (minutes/day), classification accuracy, draft acceptance rate, "missed-important" false-negative rate |
| Eval suite | 22 test cases (80% common, 15% edge, 5% adversarial, phishing + prompt injection). 97% on our internal eval (how we test). Re-runs after every prompt or knowledge-base update. |
| Audit trail | Every email classified, labelled, drafted, surfaced, or archived is logged, along with each approval. Full reproducibility for any action. |
What Lyra handles
The categories she prepares for you.
Scheduling requests
When someone asks for a meeting, Lyra checks your calendar, proposes 3 options, drafts the reply. You approve in one click.
Standard intros
Someone sends a "double opt-in intro to X"? Lyra drafts your reply pattern (acknowledge → context → next step). Faster than typing.
Vendor follow-ups
"Did you see my email last week?" → Lyra drafts the polite holding reply. "Following up on our quote" → drafts the response with current status.
FAQ patterns from your sent folder
After 30 days, Lyra identifies the 10 most common reply patterns YOU have written and starts drafting them automatically for new occurrences. You approve.
"Got it, thanks"-class
When the only correct response is acknowledgment, Lyra drafts it and queues it for your one-click approval. Once you approve, it goes out and the thread closes.
Why small models, not frontier ones
The model choice that matters at this tier.
Lyra runs on fast, low-cost models rather than a frontier model, deliberately. Task agents like Lyra do many small actions (classify 200 emails, draft 30 replies, surface 10 important items) rather than few large reasoning tasks. A small model is the right tool: faster per-call, cheaper per-action, accurate enough for bounded classification.
A frontier model would be overkill here and add unnecessary cost. A typical Lyra deployment runs €15–€40/month in API costs at founder-volume (200–300 emails/day). The same workload on a frontier model would cost €100–€300/month with no measurable accuracy improvement on her specific task profile.
This is what "right model for the tier" means in our methodology, and one reason a custom-built agent is cheaper per-action than a SaaS alternative locked into a single model choice.
FAQ
Questions about Lyra.
What does Lyra do?
AI email triage agent. Classifies inbox by urgency and topic, drafts replies to common patterns, suggests meeting times, labels newsletters and receipts, sends daily digest of items needing attention. You approve every send, booking and archive. Task tier, fast low-cost models, eval suite re-run on every change. Reclaims 60–90 minutes/day (typical estimate) for most founders.
Is Lyra a replacement for Superhuman or Shortwave?
No, complementary. Those are email clients that make YOU faster. Lyra is an agent that handles email FOR you. She sits on top of any Gmail or Outlook account; you keep your preferred client to read what she surfaces.
How does Lyra handle privacy?
OAuth scope-limited access. Model APIs configured for zero retention (no email content retained beyond 7-day operational logs, never used for training). Sensitive senders/keywords excluded by rule. Full audit trail. DPA before deployment.
Can Lyra send emails autonomously?
No. She drafts and you approve, every time. Tuning makes her drafts better and your approvals faster (one click, or a batch for simple acknowledgments and scheduling confirmations), but nothing goes out without you. The same applies to booking meetings and archiving.
What if Lyra mis-classifies an important email?
Designed to bias toward over-surfacing. The "missed-important" false-negative rate is the most-tracked KPI. If she's not sure, she surfaces it. The cost of one extra surface is 10 seconds of your attention; the cost of one miss is real consequence.
How much does customizing Lyra cost?
Lyra customized to your business: Spark (€2,500, 14 days) or part of Workforce. Operating cost €15–€40/month at typical founder volume, small models are intentionally cheap-per-action.
Does Lyra work for shared inboxes?
Yes, with one additional configuration: she needs to know which actions each team member is authorized for. Typical setup at a 3–5 person team takes 2 extra days of tuning beyond solo Spark scope.
What if I get a phishing email?
Lyra's eval suite explicitly tests for prompt-injection-via-email-content. She does not execute instructions found inside emails. Suspicious patterns (urgent + financial + new sender) escalate to you with a "potential phishing" flag rather than getting labelled as noise.
