The problem
Stitched speech-to-text, LLM and text-to-speech voice agents are slow, brittle and lose tone on real calls
What they do
Speech-to-speech voice models and an API for building production voice agents
Traction Company claims ~20x growth in 2026; customers include Ashby, Maven AGI, Rho, a Fortune 50
Hiring trend
- Open roles
- 7
- Ops roles
- 1
- Headcount
- ~10-23
Open roles 3 (Mar 2026) to 7 (Nov 2026); headcount ~23, +28% YoY (LinkedIn est.)
Hiring in
- San Francisco
Latest news
Oct 2026
Launched Phonic-1 speech-to-speech model and announced $27.5M Series A led by First Round ↗Funding
- Last round
- $27.5M
- Stage
- Series A
- Date
- Oct 2026
Led by First Round
- Hiring+17 open roles on the live job board (6–9 roles = +1)
- Funding recency+1.5Series A · $27.5M, Oct 2026 (1 month ago): 6 months or less = +1.5
- Traction+0.75Some growth: Company claims ~20x growth in 2026; customers include Ashby, Maven AGI, Rho, a Fortune 50
- Red flags0None found in the last 24 months
- Total3.251 + 1.5 + 0.75 − 0 = 3.25 → Steady (1.5–3.49)
Distance (Unproven) is a separate read on how long the lead can last. It doesn't change the score.
Distance
Unproven · edge: learns from use
Two-year-old, ~10-23 person team whose model tops independent benchmarks, but it sells straight against OpenAI, Google and xAI voice models plus ElevenLabs, Vapi and Retell, and has published no revenue or retention data.
Sources: Phonic blog: Phonic-1 launch and $27.5M Series A, compares itself to GPT-Live, Grok Voice and Gemini Live (Oct 2026) ↗ · Magica: benchmarks show Phonic-1 leads on reliability but rivals are faster or cheaper; no retention data disclosed (Oct 2026) ↗
Open ops roles · 1
Last checked Nov 4, 2026