The intake AI handles bilingual calls seamlessly and routes high-value cases to the right attorney immediately. We stopped paying for a call center and got better results. That never happens.
Or subscribe for updates
Lily founded her firm six years ago in a metro area with a large Spanish-speaking population. From day one, bilingual intake wasn't optional. It was a core requirement. About 45% of her inbound calls come in Spanish.
For the first four years, Lily handled it herself. She's fluent in both languages. But as the firm grew to three attorneys and the call volume crossed 200 inbound calls per week, she couldn't keep answering the phone between depositions.
She hired a bilingual call center. The service provided live operators who could field calls in English and Spanish, take messages, and transfer urgent calls to the office. Monthly cost: roughly $9,500.
The problems started almost immediately.
Quality was inconsistent. The call center rotated operators frequently. Some spoke excellent Spanish, others stumbled through basic legal terminology. Callers noticed. Several mentioned it in consultations, saying they'd almost hung up because the person on the phone didn't seem to understand their situation.
Routing was crude. The call center used a simple flowchart: personal injury goes to Attorney A, family law goes to Attorney B, everything else goes to voicemail. But Lily's firm has more nuanced routing needs. Cases involving commercial vehicles go to a specific attorney, high-value cases get flagged for partner review, and cases in certain counties need to be declined upfront.
Data was sparse. Each call produced a one-paragraph message. No structured data, no qualification scoring, no intake form. Lily's team spent significant time on callback calls just recollecting information the caller had already provided once.
Kenstera deployed a bilingual AI intake agent that handles both English and Spanish natively. Not through translation, but through language-specific conversational models trained on legal intake workflows.
The AI detects the caller's preferred language within the first few seconds and continues the entire conversation in that language. There's no "press 1 for English" menu. No awkward language switching. The experience feels natural because the AI isn't translating on the fly. It's conducting the conversation in the caller's language from the start.
The AI implements the firm's actual routing rules, not a simplified version:
A Spanish-speaking caller dials in at 9 PM on a Thursday after a car accident:
The entire call takes about four minutes. The intake summary, written in English for the attorney, is in the CRM before the caller hangs up.
Lily cancelled the call center contract after the first month. The AI outperformed it on every metric that matters.
Monthly cost dropped from $9,500 to approximately $1,000. That's an $8,500/month savings, over $100K annualized, for a service that actually performs better.
Caller satisfaction improved measurably. The firm surveys new clients at the end of their first consultation. Since switching to AI intake, satisfaction scores for the initial call experience went from 3.6/5 to 4.7/5. Spanish-speaking callers specifically noted that the intake call felt "like talking to someone who actually understood."
Routing accuracy hit 98%. Cases land with the right attorney on the first try. No more Monday morning reshuffling when someone realizes a family law case got sent to the PI attorney, or a commercial vehicle case wasn't flagged for partner review.
| Metric | Call Center | Kenstera AI |
|---|---|---|
| Monthly cost | $9,500 | ~$1,000 |
| Language quality (Spanish) | Inconsistent | Native-level |
| Routing accuracy | ~70% | 98% |
| Intake data completeness | 1 paragraph | Full structured form |
| Caller satisfaction | 3.6/5 | 4.7/5 |
| Average answer time | 12 seconds | 1.8 seconds |
The attorneys noticed the difference in consultation quality. When a client walks in and the attorney already has a complete intake summary with accident details, injury status, insurance information, and liability factors, the first meeting is productive from minute one. No more spending 20 minutes recollecting information.
Lily is expanding the AI's role beyond intake. The next phase includes automated appointment reminders (in the client's preferred language), post-consultation follow-up messages, and document request sequences.
She's also testing the AI for outbound follow-up, reaching out to leads who called but didn't book, offering to reschedule at a convenient time. Early results show a 23% re-engagement rate on those callbacks.
The firm hasn't hired a replacement receptionist. The AI handles the first touch, and the paralegals handle everything from there. The overhead dropped, the quality went up, and Lily stopped answering the phone during depositions. That last part, she says, might be the biggest win.